On Mon, 01 Jul 2024 15:50:30 +0200, Ahmad Fatoum wrote:
> This is a port of U-Boot commit eaf3e613ea6f0dc95c94a93997ad62785fe2969c:
>
> | Author: Julius Werner
> | Date: Fri Jul 19 13:12:08 2013 -0700
> |
> | usb: Use well-known descriptor sizes when parsing configuration
> |
> | Th
On Tue, 02 Jul 2024 08:00:25 +0300, Alexander Shiyan wrote:
> Additionally, the indentation of spaces has been replaced by tabs.
>
>
Applied, thanks!
[1/1] commands: cpuinfo: Add Cortex-A55 information
https://git.pengutronix.de/cgit/barebox/commit/?id=0ec1a2bcf2db (link may
not be sta
On Mon, 01 Jul 2024 20:12:39 +0200, Ahmad Fatoum wrote:
> GCC reports:
>
> drivers/regulator/fan53555.c:362:17: warning: ‘mode_reg’ may be used
> uninitialized [-Wmaybe-uninitialized]
> 362 | regmap_update_bits(di->rdev.regmap, mode_reg,
> mode_mask, 0);
>
> Fix this by returni
On Mon, 01 Jul 2024 20:12:30 +0200, Ahmad Fatoum wrote:
> Most of the file is macros, but the little C code there is fails to
> compile when built with !COMMON_CLK due to struct clk_div_table and
> clk_hw_get_parent being undefined.
>
>
Applied, thanks!
[1/1] clk: support inclusion of clk-pro
Additionally, the indentation of spaces has been replaced by tabs.
Signed-off-by: Alexander Shiyan
---
arch/arm/cpu/cpuinfo.c | 20
1 file changed, 12 insertions(+), 8 deletions(-)
diff --git a/arch/arm/cpu/cpuinfo.c b/arch/arm/cpu/cpuinfo.c
index 2d3fe2ac8d..ce5f010975 100
Most of the file is macros, but the little C code there is fails to
compile when built with !COMMON_CLK due to struct clk_div_table and
clk_hw_get_parent being undefined.
Signed-off-by: Ahmad Fatoum
---
include/linux/clk-provider.h | 2 ++
include/linux/clk.h | 6 ++
2 files changed
GCC reports:
drivers/regulator/fan53555.c:362:17: warning: ‘mode_reg’ may be used
uninitialized [-Wmaybe-uninitialized]
362 | regmap_update_bits(di->rdev.regmap, mode_reg, mode_mask,
0);
Fix this by returning -EINVAL in that case, which is the error code used
earlier if the vendo
This is a port of U-Boot commit eaf3e613ea6f0dc95c94a93997ad62785fe2969c:
| Author: Julius Werner
| Date: Fri Jul 19 13:12:08 2013 -0700
|
| usb: Use well-known descriptor sizes when parsing configuration
|
| The existing USB configuration parsing code relies on the descriptors'
| o
On Mon, 01 Jul 2024 09:52:32 +0200, Ahmad Fatoum wrote:
> The kernel device trees for the platform already describe the RTC, so
> let's do the same in barebox as well.
>
>
Applied, thanks!
[1/1] ARM: dts: imx8mp-skov: describe RTC
https://git.pengutronix.de/cgit/barebox/commit/?id=0f914
On Mon, 01 Jul 2024 12:08:59 +0200, Ahmad Fatoum wrote:
> The barebox area reservation is meant to fix the issue of having gaps
> between early barebox-related reservations that might be clobbered by
> later calls to request_sdram_region().
>
> The OP-TEE memory region, if it exists, is not a ba
On Mon, 01 Jul 2024 12:35:12 +0200, Ahmad Fatoum wrote:
> This function is currently only used by the sedit command, but can be
> useful for other "full-screen" commands as well.
>
>
Applied, thanks!
[1/3] commands: edit: factor out getwinsize and export it for reuse
https://git.pengutr
On Mon, 01 Jul 2024 12:22:29 +0200, Ahmad Fatoum wrote:
> Booting barebox over TFTP is a common operation that even has a default
> boot target: `boot bnet`.
>
> Once, a chainloaded barebox was tested to work, it may need to be
> persisted, which needs a cumbersome:
>
> barebox_update /mnt/tf
On Wed, 19 Jun 2024 00:33:57 +0300, Antony Pavlov wrote:
> Here is the code from drivers/spi/spi-fsl-qspi.c:
>
> > res = dev_request_mem_resource(dev, 0);
> > q->iobase = IOMEM(res->start);
> > if (IS_ERR(q->iobase)) {
> > ret = PTR_ERR(q->iobase);
> > goto err_put_ctrl;
> > }
>
On Wed, 19 Jun 2024 07:55:37 +0300, Antony Pavlov wrote:
>
Applied, thanks!
[1/1] images: gitignore more files
https://git.pengutronix.de/cgit/barebox/commit/?id=a53326440a6b (link may
not be stable)
Best regards,
--
Sascha Hauer
On Fri, Jun 21, 2024 at 08:18:06AM +0300, Alexander Shiyan wrote:
> This driver supports Fairchild FAN53555 Digitally Programmable TinyBuck
> Regulator.
>
> Supported ICs:
> fan53526, fan53555, rk8600, rk8602, syr827, syr828, tcs4525, tcs4526.
>
> The driver is written based on the fan53555.c li
On Fri, 21 Jun 2024 08:18:06 +0300, Alexander Shiyan wrote:
> This driver supports Fairchild FAN53555 Digitally Programmable TinyBuck
> Regulator.
>
> Supported ICs:
> fan53526, fan53555, rk8600, rk8602, syr827, syr828, tcs4525, tcs4526.
>
> The driver is written based on the fan53555.c linux
On Thu, 20 Jun 2024 09:16:16 +0200, Ahmad Fatoum wrote:
> Calling bootm on an x86 kernel image that lacks an EFI stub will result
> in a "no image handler found for image type MBR sector", which can be
> very confusing.
>
> Improve upon this by directly suggesting that the kernel's
> CONFIG_EFI_
On Fri, 21 Jun 2024 08:33:11 +0300, Alexander Shiyan wrote:
> The patch is part of the preparation for adding the SPI interface support.
>
>
Applied, thanks!
[1/1] mfd: rk808: Abstracting childs from the interface being used
https://git.pengutronix.de/cgit/barebox/commit/?id=ba02270e1fc
On Thu, 20 Jun 2024 10:49:20 +0300, Alexander Shiyan wrote:
>
Applied, thanks!
[1/1] mfd: rk808: Fix poweroff handler name
https://git.pengutronix.de/cgit/barebox/commit/?id=af28a9d08323 (link may
not be stable)
Best regards,
--
Sascha Hauer
On Sat, 22 Jun 2024 19:48:14 +0200, Oleksij Rempel wrote:
> Current implementation of GPIO-based chip select configuration in this
> driver only considers the first valid GPIO and ignores
> subsequent entries. This patch ensures that all valid GPIO entries are
> processed correctly, enabling conf
Hi Oleksij,
Some small nitpicks inside.
On Sat, Jun 22, 2024 at 07:48:15PM +0200, Oleksij Rempel wrote:
> +++ b/arch/arm/boards/microchip-sama5d3-eds/board.c
> @@ -0,0 +1,414 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +// SPDX-FileCopyrightText: 2024 Oleksij Rempel
> +
> +#include
> +
On Sun, 23 Jun 2024 23:46:53 +0300, Denis Orlov wrote:
> Switch from using u32 to unsigned long for these. It is unlikely to be
> actually utilized on 64-bit builds, but this feels a bit safer,
> considering that we work with the whole register width (32 or 64 bit,
> depending on the config setti
On Thu, 20 Jun 2024 14:58:29 +0200, Ahmad Fatoum wrote:
> When panic() is not configured to hang, it will wait 100 ms for messages
> to go out before trying to poweroff or reset the machine.
>
> The wait is implemented using udelay(), which runs pollers if the delay
> duration exceeds 100 us. Th
On Thu, 20 Jun 2024 14:55:47 +0200, Ahmad Fatoum wrote:
> base: class: fix class_remove_device to not corrupt the class list
>
> class_remove_device is supposed to remove the device from the class, not
> remove the class from its devices.
>
>
Applied, thanks!
[1/2] fixup! base: add class dev
On Wed, Jun 26, 2024 at 12:04:42PM +0200, Marco Felsch wrote:
> On 24-06-17, Sascha Hauer wrote:
> > On Tue, Jun 11, 2024 at 10:36:47AM +0200, Marco Felsch wrote:
> > > Hi,
> > >
> > > sorry for the delay on this patchset.
> > >
> > > On 24-03-25, Sascha Hauer wrote:
> > > > Hi Marco,
> > > >
>
For testing proper operation of IIO devices, it can be useful to monitor
changes in the reading reported by the hwmon command. This is now
possible by using `watch -n 0.5 hwmon`.
Signed-off-by: Ahmad Fatoum
---
v1 -> v2:
- return success when command is interrupted (Sascha)
- free all allocat
We have no floating point support in barebox and no strtod. Still some
user interaction could benefit from being able to supply command
arguments or device parameters in a format with decimal points.
This new simple_strtofract() function parses a fixed point format consisting
of a 32 bit integer a
This function is currently only used by the sedit command, but can be
useful for other "full-screen" commands as well.
Signed-off-by: Ahmad Fatoum
---
v1 -> v2:
- no change
---
commands/edit.c | 85 +++--
include/term.h | 8 +
lib/Makefile|
On Mon, 01 Jul 2024 08:14:19 +0200, Steffen Trumtrar wrote:
> The i.MX93 needs to set the RMII speed to successfully reset. Add a
> callback that does this inspired by the linux patch:
>
> net: stmmac: dwmac-imx: use platform specific reset for imx93 SoCs
> commit b536f32b5b034f592df0f0b
On Mon, 01 Jul 2024 09:04:12 +0200, Ahmad Fatoum wrote:
> Import the Linux v6.10-rc3 state of the RTC/NVMEM driver, so it can be
> used to check wall time in barebox.
>
>
Applied, thanks!
[1/1] rtc: add PCF85063 RTC and NVMEM support
https://git.pengutronix.de/cgit/barebox/commit/?id=1f
Booting barebox over TFTP is a common operation that even has a default
boot target: `boot bnet`.
Once, a chainloaded barebox was tested to work, it may need to be
persisted, which needs a cumbersome:
barebox_update /mnt/tftp/${global.user}-barebox-${global.hostname}
Make this more straight-fo
Hello Sascha,
On 01.07.24 12:08, Sascha Hauer wrote:
> On Mon, Jul 01, 2024 at 09:14:55AM +0200, Ahmad Fatoum wrote:
>> Booting barebox over TFTP is a common operation that even has a default
>> boot target: `boot bnet`.
>>
>> Once, a chainloaded barebox was tested to work, it may need to be
>> pe
On Mon, 01 Jul 2024 09:05:47 +0200, Ahmad Fatoum wrote:
> A failure in persistent_ram_init_ecc() ultimately results in
> ramoops_probe() failing, so an error message is more appropriate than
> the current info log level.
>
>
Applied, thanks!
[1/1] fs: pstore: ram_core: use pr_err, not pr_info
On Mon, 01 Jul 2024 09:07:24 +0200, Ahmad Fatoum wrote:
> We actually never check the device parameter points again in the MCI
> core, so we don' need to keep them in struct mci.
>
>
Applied, thanks!
[1/1] mci: core: drop unused parameter members from struct mci
https://git.pengutronix.
On Mon, 01 Jul 2024 09:12:22 +0200, Ahmad Fatoum wrote:
> The GFP arguments are unused in barebox and are only used to reduce
> churn when porting kernel code. To make it easier to port code using
> them in headers, move them to like Linux does.
>
>
Applied, thanks!
[1/2] include: gfp: move
The barebox area reservation is meant to fix the issue of having gaps
between early barebox-related reservations that might be clobbered by
later calls to request_sdram_region().
The OP-TEE memory region, if it exists, is not a barebox-related
reservation and treating it as such will clash if the
On Mon, Jul 01, 2024 at 09:14:55AM +0200, Ahmad Fatoum wrote:
> Booting barebox over TFTP is a common operation that even has a default
> boot target: `boot bnet`.
>
> Once, a chainloaded barebox was tested to work, it may need to be
> persisted, which needs a cumbersome:
>
> barebox_update /mn
On Mon, 01 Jul 2024 09:27:14 +0200, Ahmad Fatoum wrote:
> Despite the backlight parameter being registered as uint32,
> backlight_brightness_set took the brightness level as a signed value.
>
> This can lead to wrap around for very large values of brightness
> supplied as device parameter and if
On Mon, 01 Jul 2024 09:28:50 +0200, Ahmad Fatoum wrote:
> driver: wrap class macro arguments in parenthesis
>
> Without the parenthesis, using & in the macro argument will lead to
> issues.
>
>
Applied, thanks!
[1/1] fixup! base: add class device support
https://git.pengutronix.de/cgit
On Mon, 01 Jul 2024 09:32:13 +0200, Ahmad Fatoum wrote:
> The Rockchip pinctrl binding can not only mux pins as GPIOs and
> configure the bias, but also configure the direction and whether it's
> high/low for outputs and schmitt trigger for inputs.
>
> This is used in some device trees to enable
On Mon, 01 Jul 2024 09:35:09 +0200, Ahmad Fatoum wrote:
> test_bit currently returns some positive integer on success and don't
> enforce the return values to be either false or true.
>
> This can lead to problems if the return value is not used for
> truthiness, but in arithmetic.
>
> This see
On Mon, 01 Jul 2024 09:35:24 +0200, Ahmad Fatoum wrote:
> The invm device is specifically created for the EEPROM, so adding a
> parameter should not result in a name clash.
>
> This leaves only -ENOMEM as possible error. As the driver is usable
> without EEPROM being writable, don't propagate th
On Mon, 01 Jul 2024 09:38:22 +0200, Ahmad Fatoum wrote:
> This introduces no functional change, but aligns us with the coding
> style, where early exits are preferred and the final return is the
> successful one.
>
>
Applied, thanks!
[1/1] of: platform: early exit on error
https://git.p
On Mon, 01 Jul 2024 09:43:25 +0200, Ahmad Fatoum wrote:
> We already have an RTC class device, so by using struct class, we can
> drop the list member in struct rtc_device and make it possible to list
> the device using the class command.
>
>
Applied, thanks!
[1/1] rtc: register RTC class to
On Mon, Jul 01, 2024 at 09:54:34AM +0200, Ahmad Fatoum wrote:
> For testing proper operation of IIO devices, it can be useful to monitor
> changes in the reading reported by the hwmon command. This is now
> possible by using `watch -n 0.5 -t hwmon`.
>
> Signed-off-by: Ahmad Fatoum
> ---
> comman
On Mon, 01 Jul 2024 10:48:25 +0200, Ahmad Fatoum wrote:
> ChatGPT can explain ASCII art, but without information about the
> position, it's not as useful for a blind user.
>
> Extend the docs, so it's more useful in that case.
>
>
Applied, thanks!
[1/1] Documentation: boards: stm32mp: descri
ChatGPT can explain ASCII art, but without information about the
position, it's not as useful for a blind user.
Extend the docs, so it's more useful in that case.
Signed-off-by: Ahmad Fatoum
---
Documentation/boards/stm32mp.rst | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff -
For testing proper operation of IIO devices, it can be useful to monitor
changes in the reading reported by the hwmon command. This is now
possible by using `watch -n 0.5 -t hwmon`.
Signed-off-by: Ahmad Fatoum
---
commands/Kconfig | 6 +++
commands/Makefile | 1 +
commands/watch.c | 97 +
We have no floating point support in barebox and no strtod. Still some
user interaction could benefit from being able to supply command
arguments or device parameters in a format with decimal points.
This new simple_strtofract() function parses a fixed point format consisting
of a 32 bit integer a
This function is currently only used by the sedit command, but can be
useful for other "full-screen" commands as well.
Signed-off-by: Ahmad Fatoum
---
commands/edit.c | 85 +++--
include/term.h | 8 +
lib/Makefile| 1 +
lib/term.c | 69
For testing proper operation of IIO devices, it can be useful to monitor
changes in the reading reported by the hwmon command. This is now
possible by using e.g. `watch -n 0.5 -t hwmon`.
Ahmad Fatoum (3):
commands: edit: factor out getwinsize and export it for reuse
lib: strtox: implement new
The kernel device trees for the platform already describe the RTC, so
let's do the same in barebox as well.
Signed-off-by: Ahmad Fatoum
---
v1 -> v2:
- fix pinmux broken by last-minute I2C GPIO change..
---
arch/arm/dts/imx8mp-skov.dts | 40
1 file changed,
We already have an RTC class device, so by using struct class, we can
drop the list member in struct rtc_device and make it possible to list
the device using the class command.
Signed-off-by: Ahmad Fatoum
---
drivers/rtc/class.c | 10 ++
include/linux/rtc.h | 1 -
2 files changed, 6 ins
This introduces no functional change, but aligns us with the coding
style, where early exits are preferred and the final return is the
successful one.
Signed-off-by: Ahmad Fatoum
---
drivers/of/platform.c | 15 ---
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/drivers
The invm device is specifically created for the EEPROM, so adding a
parameter should not result in a name clash.
This leaves only -ENOMEM as possible error. As the driver is usable
without EEPROM being writable, don't propagate the error.
Signed-off-by: Ahmad Fatoum
---
drivers/net/e1000/mtd.c
The bridge device is specifically created for purposes of the FPGA
bridge framework. There should thus be no naming clashes and
-ENOMEM is not a fatal error there either, because the PWM C API
is still usable without the device parameters.
Signed-off-by: Ahmad Fatoum
---
drivers/fpga/fpga-bridge
The PWM device is specifically created for purposes of the PWM
framework. There should thus be no naming clashes and -ENOMEM
is not a fatal error there either, because the PWM C API is
still usable without the device parameters.
Signed-off-by: Ahmad Fatoum
---
drivers/pwm/core.c | 25 +--
test_bit currently returns some positive integer on success and don't
enforce the return values to be either false or true.
This can lead to problems if the return value is not used for
truthiness, but in arithmetic.
This seems to be the case only in KASAN code, where it's used to count
errors. C
The older rk3128, rk3036 and rk3066a SoCs only supported a
pin-function-specific default bias as well as disabling bias with no way
to manually choose whether the bias should be a pull-up or a pull-down.
We wrongly treated this as if there is no bias, so let's fix that.
Signed-off-by: Ahmad Fatou
The Rockchip pinctrl binding can not only mux pins as GPIOs and
configure the bias, but also configure the direction and whether it's
high/low for outputs and schmitt trigger for inputs.
This is used in some device trees to enable peripherals as a finer
grained gpio-hog.
This series adds that sup
The header is an integral part of the Linux pinctrl support, because
pinctrl drivers parse the device tree into the packed format described
by it and receive the same format later when a pinctrl state needs to be
selected.
We don't follow this same procedure in barebox and instead parse pinctrl
en
No Rockchip device tree actually sets a value for these properties and
neither kernel or barebox driver cares about the property value.
Thus replace of_property_read_u32 with of_property_read_bool for
simplification.
Signed-off-by: Ahmad Fatoum
---
drivers/pinctrl/pinctrl-rockchip.c | 8 +++
The Rockchip pinctrl binding can not only mux pins as GPIOs and
configure the bias, but also configure the direction.
This is used in some device trees to enable peripherals as a finer
grained gpio-hog.
In Linux, this is implemented by keeping a list of deferred pin configs,
but for barebox, let'
So far we defined our own enum to record what bias was indicated by the
device tree properties. Now that we have imported the upstream enum
definition, make use of it to make syncs with Linux easier.
While at it, we replace one usage of RK_BIAS_PULL_UP with `1'. That they
were equal before was a c
Rockchip pinctrl binding can configure GPIO direction, so let's add a
function that returns a gpio_chip that can be used for this.
Signed-off-by: Ahmad Fatoum
---
drivers/gpio/gpiolib.c | 19 +++
include/gpio.h | 6 ++
2 files changed, 25 insertions(+)
diff --git a/
We had code to calculate the register layout for setting schmitt
triggers on pins, but it was not actually used.
Fix this and start parsing the input-schmitt-enable/disable properties.
Signed-off-by: Ahmad Fatoum
---
drivers/pinctrl/pinctrl-rockchip.c | 40 ++
1 file
driver: wrap class macro arguments in parenthesis
Without the parenthesis, using & in the macro argument will lead to
issues.
Signed-off-by: Ahmad Fatoum
---
include/device.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/include/device.h b/include/device.h
index d8d1f7
We already emulate 64KiB of port space by directing (in|out)[bwl] accesses
to the __pci_iobase static array.
By have (read|write)[bwlq] access the first 4K of that port space when
trying to access the NULL page.
That way generic drivers like e.g. pinctrl-single can be trivially
described in the d
Now that MMIO addresses 0-1023 point at valid memory, let's use it
describe a fake pinctrl-single controller. This allows testing common
pinctrl code in sandbox.
Device parameter support to select the state will follow later.
Signed-off-by: Ahmad Fatoum
---
arch/sandbox/dts/sandbox.dts | 53 +++
Despite the backlight parameter being registered as uint32,
backlight_brightness_set took the brightness level as a signed value.
This can lead to wrap around for very large values of brightness
supplied as device parameter and if this results in a negative
brightness, the function appears to hang
The kernel device trees for the platform already describe the RTC, so
let's do the same in barebox as well.
Signed-off-by: Ahmad Fatoum
---
arch/arm/dts/imx8mp-skov.dts | 39
1 file changed, 39 insertions(+)
diff --git a/arch/arm/dts/imx8mp-skov.dts b/arch/a
Booting barebox over TFTP is a common operation that even has a default
boot target: `boot bnet`.
Once, a chainloaded barebox was tested to work, it may need to be
persisted, which needs a cumbersome:
barebox_update /mnt/tftp/${global.user}-barebox-${global.hostname}
Make this more straight-fo
Including all of common.h just to get the barebox version is overkill.
Let's move the version/info declarations into a separate header that can
be used instead.
Signed-off-by: Ahmad Fatoum
---
include/barebox-info.h | 31 +++
include/common.h | 23 +-
The GFP arguments are unused in barebox and are only used to reduce
churn when porting kernel code. To make it easier to port code using
them in headers, move them to like Linux does.
Signed-off-by: Ahmad Fatoum
---
include/linux/gfp.h | 12
include/linux/slab.h | 7 +--
2 fi
We actually never check the device parameter points again in the MCI
core, so we don' need to keep them in struct mci.
Signed-off-by: Ahmad Fatoum
---
drivers/mci/mci-core.c | 14 ++
include/mci.h | 2 --
2 files changed, 6 insertions(+), 10 deletions(-)
diff --git a/drive
A failure in persistent_ram_init_ecc() ultimately results in
ramoops_probe() failing, so an error message is more appropriate than
the current info log level.
Signed-off-by: Ahmad Fatoum
---
fs/pstore/ram_core.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/pstore/ram_co
Import the Linux v6.10-rc3 state of the RTC/NVMEM driver, so it can be
used to check wall time in barebox.
Signed-off-by: Ahmad Fatoum
---
drivers/rtc/Kconfig| 14 ++
drivers/rtc/Makefile | 2 +
drivers/rtc/nvmem.c| 29 +++
drivers/rtc/rtc-pcf85063.c | 473 +
77 matches
Mail list logo
